Adiponectin translation is increased by the PPARγ agonists pioglitazone and ω-3 fatty acids

Animals.

Male Sprague-Dawley rats (300 g) were purchased from Harlan Sprague Dawley (Indianapolis, IN). Animal care was performed in accordance with Institutional Animal Care and Use Committee (IACUC) guidelines. The study was approved by the IACUC at the Central Arkansas Veterans Healthcare System, Little Rock, AR, protocol no. 2-05-1. Adipocytes were prepared from epididymal fat pads, as described previously (27, 30).

Cell culture and maintenance.

3T3-F442A cells were obtained from Dr. Howard Green (Harvard Medical School, Boston, MA). For experiments, cells were grown to confluence and stimulated to differentiate in DMEM containing 10% fetal calf serum and 100 nM insulin for 14 days (3).

Preparation of fatty acids.

DHA (22:6, n-3), EPA (20:5, n-3) (Sigma), palmitic acid, and oleic acid (MP Biomedicals) were dissolved in ethanol and conjugated to fatty acid-free BSA in a 2.5:1 ratio at 45°C for 30 min. The adipocytes were treated with 25 μM of each of the fatty acids or 3 μM pioglitazone in presence or absence of 1.5 μM PPARγ inhibitor, GW-9662 (Sigma) or 1 mM of hydroxylysine glycosylation inhibitor, and 2,2′-dipyridyl (Cole-Parmer). The ethanol concentration of the medium was always <0.0002%.

Western blots for adiponectin.

Adipocytes (150 μl) were incubated in DMEM-HEPES buffer supplemented with 1% FBS and pen-strep containing pioglitazone or conjugated fatty acid. Cells were lysed in RIPA lysis buffer containing protease inhibitor and analyzed on 4–20% Criterion precast gels (Bio-Rad), followed by Western blotting with goat anti-mouse adiponectin antibody (R & D Systems). The different isoforms were quantified and analyzed as a percent of total adiponectin.

Metabolic labeling of adipocytes and immunoprecipitation.

Rat adipocytes were incubated overnight, as described above, in the presence of either pioglitazone or the different fatty acids. On the following day, cells were starved for 1 h in methionine-free DMEM and pulse labeled for 15 min with medium containing 100 μCi/ml [³⁵S]methionine. The labeling medium was replaced with DMEM, and equal amounts of adipocytes were pulse chased for the indicated time intervals by incubating at 37°C. Cells and medium were separated; adipocytes were lysed in 1 ml of RIPA lysis buffer with PMSF (1 μM). Cell lysates and medium were precleared with protein A/G beads (Roche), and supernatants were immunoprecipitated with goat anti-mouse adiponectin IgG. Immunoprecipitated complexes were separated on 10% SDS-PAGE reducing gels followed by autoradiography. The radioactivity associated with each band was quantified with the Image Quant TL software (Amersham Biosciences). At the end of the 15-min pulse, 10% TCA precipitable counts was determined using 10 μl from the total cell lysates.

Total RNA isolation and real-time RT-PCR.

RNA isolation from adipocytes and real-time PCR was performed as described earlier (22). Primer sequences for rat adiponectin RNA are described below, and data are expressed in relation to 18S RNA. The data represent arbitrary units that accurately compare the samples with each other within that assay. The threshold cycle values of the PCR reactions were generally between 20 and 30 for all assays.

Primers were rat adiponectin: 5′-TAAGGGTGACCCAGGAGATG-3′(forward), 5′-GGA ACATTGGGGACAGTGAC-3′(reverse); and 18S: 5′-TTCGAACGTCTGC-3′ (forward), 5′-ATGGTAGGCACGGCGACTA-3′ (reverse).

Preparation of S-100 extracts.

Cytoplasmic extract was made from adipocytes following treatment with pioglitazone or DHA. Approximately 2 ml of rat adipocytes or ∼2.5 × 10⁵ 3T3-F442A adipocytes were used for the preparation of S-100 extracts. A cytoplasmic extract was prepared as described previously (45). The postnuclear extract was used to prepare a high-speed supernatant fraction (S-100); total protein was estimated in the S-100 extracts, and equal amounts of protein were compared using in vitro translation assay.

Adiponectin construct used for in vitro translation.

Rat adiponectin clone was purchased from ATCC (NM_144744). This clone contains 940 nucleotides; this includes 21 bases of 5′-untranslated region (UTR), the complete coding sequence nt 22–756, and 184 bases of 3′-UTR cloned in a transcription vector. The plasmid was linearized, and transcripts were generated using the SP6 RNA polymerase.

In vitro translation.

In vitro translation of RNA transcripts was performed as described previously (45). Equal quantities of RNA transcripts (0.2 μg) were translated in a rabbit reticulocyte lysate system (Promega) in the presence of [³⁵S]methionine. Reactions were incubated at 30°C for 1 h and terminated by transferring to 4°C. The products of reactions were analyzed on 10% SDS-PAGE followed by autoradiography. Images were quantitated using Image Quant TL software (Amersham Biosciences).

Polysome preparation and real-time PCR.

Polysome profiles were obtained as described previously (20, 28, 45). In brief, postmitochondrial supernatants were prepared from rat adipocytes treated overnight with pioglitazone (3 μM), DHA (25 μM), or vehicle alone (control). The adipocytes were washed once in PBS and homogenized in 2 vol of buffer I (10 mM Tris·HCl, pH 7.4, 10 mM KCl, 5 mM MgCl₂, 20 U/ml RNasin, 2.75 mM dithiothreitol, 0.1% Triton X-100, 150 μg/ml cycloheximide, 250 μg/ml heparin), using 10 strokes in a Dounce homogenizer. Mitochondria and nuclei were pelleted at 10,000 g for 10 min, and the postmitochondrial supernatant was reconstituted with 0.25 M sucrose and 0.1 M KC1 and layered over 10–50% sucrose gradients prepared in buffer II (20 mM HEPES, pH 7.2, 0.25 M KCl, 10 mM MgCl₂, 20 mM dithiothreitol, 150 μg/ml cycloheximide, 100 U/ml RNasin, and 0.5 μg/ml heparin). The gradients were centrifuged at 180,000 g for 3 h at 4°C. Fractions were collected from the top in 17 750-μl fractions, and polysome profiles were recorded by reading UV absorption at 260 nm. Each fraction was precipitated, and RNA was extracted. Measurement of adiponectin mRNA levels in each fraction was done using real-time PCR. The primers for this reaction were derived from the rat adiponectin cDNA sequence.

Statistics.

Data are expressed as means ± SE, and Student's t-test is used for statistical analysis. Level of statistical significance is set at P ≤ 0.05. For multiple comparisons, ANOVA was used with post hoc t-tests.

Alteration in adiponectin expression following treatment of rat adipocytes with pioglitazone, EPA, or DHA.

Previous studies have demonstrated that pioglitazone increases the secretion of HMW adiponectin in adipocytes (1). To better understand the mechanism of the increased adiponectin secretion, rat adipocytes were treated with pioglitazone (3 μM), DHA (25 μM), or EPA (25 μM) for 4 or 24 h, and cell extracts and medium were analyzed by Western blotting using nondenaturing gradient gels. As shown in Fig. 1A, three prominent forms of adiponectin were identified, corresponding to the HMW, MMW, and LMW forms. There was an increase in adiponectin protein in cells following treatment with pioglitazone, EPA, or DHA. At both 4 and 24 h, pioglitazone and ω-3 fatty acid treatment resulted in increased total cellular adiponectin, including all three adiponectin forms. At 4 h, cellular adiponectin was higher than at 24 h, when it was detectable in the medium. The increase in HMW was not apparent at 4 h but was evident at 24 h (Fig. 1A). There is no apparent increase in HMW adiponectin between 4 and 24 h in control medium, and a greater proportion of adiponectin was found in the LMW fraction at 24 h, suggesting some degree of degradation. The densitometric analysis of the relative expression of both total adiponectin and HMW adiponectin at 4 h in cells and 24 h in medium is illustrated in Fig. 1B. In cells, total adiponectin and HMW adiponectin increased 2.5- to threefold (P < 0.02 vs. control HMW, P < 0.05 vs. control total adiponectin) with all three treatments, pioglitazone, DHA, and EPA, whereas palmitate treatment had no effect.

In medium, both total and HMW adiponectin increased twofold with pioglitazone, DHA, and EPA, whereas palmitate had no effect. To further establish the viability of adipocytes at 4 and 24 h, cell lysates were Western blotted for lipoprotein lipase, and there was no change in lipoprotein lipase protein during this time period (data not shown).

Plasma adiponectin is increased following treatment with TZD PPARγ agonists and also by treatment with ω-3 fatty acids, apparently also through a PPARγ effect (16). However, a corresponding increase in adiponectin mRNA expression has not been observed in all instances (5, 16). In previous studies, we found no increase in adiponectin mRNA following pioglitazone treatment of human subjects (24). In the current study, adiponectin mRNA was measured in the adipocytes treated with pioglitazone, DHA, and EPA, and no changes were detected, as shown in Fig. 1C.

Pioglitazone increases the synthesis and secretion of adiponectin in rat adipocytes.

To measure adiponectin synthesis and secretion in rat adipocytes, cells were treated for 18 h with pioglitazone and then pulse labeled with [³⁵S]methionine for 15 min, followed by a chase with unlabeled medium for the indicated times (Fig. 2A). Cells and medium were separated, and adiponectin was immunoprecipitated and analyzed by SDS-PAGE and autoradiography. Under such conditions, only the 30-kDa adiponectin monomer was identified on the gel. As shown in Fig. 2A, adipocytes synthesized abundant adiponectin. When comparing control cells to pioglitazone-treated cells, at time points 0, 0.5, and 2 h after the pulse there was a 75 ± 5, 72 ± 6, and 68 ± 9% increase, respectively, in cellular adiponectin after pioglitazone treatment in four similar experiments, indicating increased synthesis of adiponectin (P < 0.02 vs. control). Total TCA-precipitable counts were determined in cell lysates from control and pioglitazone-treated cells at 0 h of chase. There was no significant change in total TCA-precipitable counts between control and pioglitazone-treated adipocytes, indicating that there was no increase in total protein synthesis. As shown in Fig. 2B, newly synthesized adiponectin was immunoprecipitated from the medium, and the appearance of newly synthesized adiponectin in the medium was time delayed compared with cellular adiponectin. During the chase, labeled adiponectin in the cells decreased at 6 h of chase, in part due to the appearance of labeled adiponectin in the medium (Fig. 2B). In the presence of pioglitazone, there was a more rapid appearance of labeled adiponectin in the medium resulting from the increased synthesis. At 24 h of chase, almost all of the labeled adiponectin was secreted into the medium, and there was a 2.8 ± 0.32-fold (P < 0.02) increase in labeled adiponectin in the medium of pioglitazone-treated cells in four independent experiments. Fold increase was calculated by densitometric analysis of autoradiograms and normalizing the data to 0 time of control.

Adiponectin synthesis and secretion: effects of ω-3 fatty acid and PPARγ in rat adipocytes.

To better assess adiponectin translation in response to PPARγ agonists, rat adipocytes were incubated overnight with pioglitazone, EPA, or DHA in the presence and absence of the PPARγ antagonist GW-9662 and then pulse labeled for 15 min with [³⁵S]methionine. At the end of the 15-min labeling, cells were isolated and 30-kDa adiponectin was immunoprecipitated and analyzed as described above. As shown in Fig. 3A, adiponectin synthesis was increased by 70 ± 5, 80 ± 5, and 60 ± 5% following treatment with pioglitazone, EPA, or DHA, respectively. However, the increase in synthesis was blocked by the addition of 1.5 μM GW-9662, indicating that the increase in adiponectin by pioglitazone, as well as by ω-3 fatty acids, was mediated by PPARγ. There was no significant change in total TCA-precipitable counts in adipocytes treated with pioglitazone or DHA in the presence or absence of GW-9662, indicating that there was no change in overall protein synthesis (data not shown). In a similar fashion, adiponectin secretion was examined in the medium (Fig. 3B). Because adiponectin secretion is time delayed compared with cellular adiponectin, cells were labeled with [³⁵S]methionine for 15 min and medium was collected after 18 h of chase. Adiponectin secretion into the medium was increased 2.8-, 3.2-, and 2.3-fold by pioglitazone, EPA, and DHA, respectively, and the presence of the PPARγ antagonist GW-9662 decreased synthesis and blocked the increase in secretion (Fig. 3B). Essentially all of the labeled adiponectin was found in the cells and the medium, and therefore, the rate of degradation was low and was unchanged by pioglitazone, DHA, or EPA.

Adiponectin synthesis and processing in rat adipocytes.

Adiponectin undergoes posttranslational modification prior to secretion involving the hydroxylation of multiple conserved proline and lysine residues and glycosylation of hydroxylysine (26). To study the importance of lysine-linked glycosylation on the synthesis of adiponectin, freshly isolated adipocytes (150 μl) were incubated with 1 mM 2,2′-dipyridyl, an inhibitor of prolyl and lysyl hydroxylases. After 3 h of glycosylation inhibition, the cells were pulse labeled for 15 min with [³⁵S]methionine followed by a chase with unlabeled medium for the indicated time period. Equal quantities of labeled adipocytes were immunoprecipitated and analyzed on SDS-PAGE followed by autoradiography. Prior to immunoprecipitation with adiponectin antibody, labeled lysates were analyzed for total protein synthesis using a small fraction of the samples to estimate TCA-precipitable counts. There was no change in TCA-precipitable counts detected in the presence or absence of 2,2′-dipyridyl (data not shown), indicating that there was no change in protein synthesis. Because this experiment involved freshly isolated adipocytes, the incorporation of [³⁵S]methionine into total TCA-precipitable counts was higher compared with adipocytes labeled the following day (Fig. 2). As shown in Fig. 4A, adiponectin was synthesized and processed rapidly. In the absence of 2,2′-dipyridyl, at least three glycosylated intermediates of adiponectin were evident in both the cells and the medium, and the increasing glycosylation and complexity of the protein was evident during the chase. When the pulse chase was conducted in the presence of 2,2′-dipyridyl, only the 30-kDa monomer was detected in the cells, and this same unglycosylated adiponectin form was secreted into the medium.

To determine whether inhibition of glycosylation would affect the formation of HMW adiponectin in the presence of pioglitazone, EPA, and DHA, adipocytes were incubated in the presence or absence of 2,2′-dipyridyl for 24 h, and cellular adiponectin was then analyzed by a nondenaturing gel and Western blot. As shown in Fig. 4B, control adipocytes contained predominantly MMW (hexameric) adiponectin, and the addition of pioglitazone, EPA, and DHA resulted in an increase in HMW adiponectin. The addition of 2,2′-dipyridyl inhibited the assembly of HMW adiponectin and resulted in a large increase in LMW, or trimeric, adiponectin. The increase in LMW adiponectin was especially pronounced in the presence of pioglitazone, EPA, and DHA, all of which increase adiponectin synthesis.

Synthetic rate of adiponectin following treatment with pioglitazone or ω-3 fatty acid in 3T3-F442A adipocytes.

To determine whether PPARγ agonists would affect adiponectin synthesis in other adipocytes, 3T3-F442A adipocytes were treated for 16 h with pioglitazone, EPA, or DHA. Adiponectin synthesis and mRNA expression were studied as described in methods. There was no change in total TCA-precipitable counts after pioglitazone, EPA, or DHA treatments, indicating that there was no change in total protein synthesis. As with rat adipocyte cultures, adiponectin synthetic rate was increased 2 ± 0.2-fold following treatment with pioglitazone and 2.8 ± 0.7- and 2.5 ± 0.4-fold with EPA and DHA treatments, respectively (Fig. 5A), whereas adiponectin mRNA expression was not changed following treatment with pioglitazone, DHA, or EPA (Fig. 5B).

Translational regulation of adiponectin.

The above data demonstrate that adiponectin synthesis was increased by pioglitazone and ω-3 fatty acids in adipocytes with no corresponding increase in mRNA. These observations would suggest regulation at the level of translation, which can be due to the presence of a cytoplasmic RNA-binding protein that specifically binds the adiponectin mRNA. To determine whether such cytoplasmic factors were present, S-100 extracts were prepared from rat adipocytes, as well as 3T3-F442A adipocyte cultures, treated for 24 h with pioglitazone or DHA. Equal amounts of S-100 cytoplasmic protein (1.25 μg) were added to a rabbit reticulocyte in vitro translation reaction containing adiponectin mRNA and labeled [³⁵S]methionine, and the products of in vitro translation were analyzed on SDS-PAGE and autoradiography. In vitro translation in the presence of the cytoplasmic extracts was compared with the addition of no cell extract (only buffer to equalize volumes). As shown in Fig. 6, the addition of cytoplasmic extracts from control (untreated adipocytes) resulted in an in vitro translation of adiponectin less robust than the addition of no cell extract. This inhibition of adiponectin translation by control cell extract was evident in both 3T3-F442A adipocytes and primary cultures of rat adipocytes (Fig. 6, A and B). The control extract from rat adipocytes did not inhibit the translation of an irrelevant mRNA (Fig. 6C). However, cell extracts from 3T3-F442A or rat adipocytes treated with pioglitazone or DHA for 24 h yielded higher levels of adiponectin translation compared with the control cell extract (Fig. 6, A and B), and these extracts did not alter translation of the irrelevant mRNA (Fig. 6C). Densitometric quantitation of the autoradiograms showed that translation was inhibited 40 ± 3% in 3T3-F442A adipocytes (Fig. 6A) and 45 ± 5% in the control rat adipocyte extracts (Fig. 6B) compared with the addition of no extract. Translation was increased by 55 ± 5% in the presence of pioglitazone- or DHA-treated adipocyte extracts compared with control in both 3T3-F442A and rat adipocytes (Fig. 6, A and B). These data suggest that adipocytes express a transacting protein that constitutively inhibits adiponectin synthesis and does not effect translation of the irrelevant mRNA used. Treatment of adipocytes with a PPARγ agonist decreases the activity of this constitutively expressed inhibitory factor, resulting in an increase in adiponectin translation.

Distribution of adiponectin mRNA on the polysome in rat adipocytes.

Earlier studies have described the regulation of translation by cytoplasmic factors that alter the association of mRNA with the polysomes (8–11). To determine whether PPARγ agonists affected initiation, we compared the distribution of adiponectin mRNA on the polysomes in control and pioglitazone- (3 μM) and DHA-treated (25 μM) adipocytes. A postmitochondrial supernatant was separated on a 10–50% sucrose gradient. Pioglitazone or DHA treatments did not change the pattern of UV absorption at 260 nm or the distribution of 18- and 28-s ribosomal RNA that was found in fractions 4–17 in all three gradients. Fractions 1–5 at the top of the gradient represent the monosomes and free mRNA, whereas the heavier fractions contain the small and larger polysomes. As shown in Fig. 7, adiponectin mRNA was distributed over a wide range of fractions. In adipocytes treated with pioglitazone or DHA, 60% of adiponectin mRNA was found in the heavier polyribosome fractions (fractions 10–17), whereas only 30% of adiponectin mRNA was present in the heavier fractions in control adipocytes (fractions 10–17), suggesting that pioglitazone and DHA treatments increased the translational initiation of adiponectin mRNA. The shift in adiponectin polysome profile tended to be stronger with DHA, which could be due to the concentration used or to other pleomorphic effects.